Cloud assessment: what to check in an AWS, Azure or GCP environment

Architecture, identity, backup, CI/CD, observability and costs: the questions by role, the evidence and the warning signs of an assessment of a cloud environment.

ProcesscloudassessmentinfrastrutturaFinOps19 November 2026 · 3 min read

A cloud environment born in a hurry almost always looks the same: resources created by hand by people who have since changed role, access keys nobody rotates, a monthly bill that grows and nobody knows who to attribute it to. A cloud assessment is for rebuilding what is there, who governs it and how much it costs to keep it like this.

The scope

It applies to workloads on AWS, Azure, GCP or equivalent platforms, to multi-cloud strategies, to migration programmes and to managed services. The first step is to understand the adoption model: lift-and-shift (move as is), re-platform (adapt) or re-architect (redesign). Everything that follows changes with it: a lift-and-shift has different cost and security problems from a native architecture.

The areas

  1. Architecture and provisioning. How the infrastructure is created. Whether Infrastructure as Code (Terraform, CloudFormation, Bicep) is used and with what coverage. What the network looks like (VPC, peering, hub-spoke, landing zone).
  2. Identity and access. How privileged accounts and service identities (IAM, managed identity, service account) are managed. Whether MFA is enforced on the console.
  3. Continuity. Backup and disaster recovery strategy, and whether the recovery objectives (RTO and RPO) have been tested recently.
  4. Operations. CI/CD, secret management (Vault, KMS, parameter store), container orchestration (EKS, AKS, GKE), observability, the patch cycle for images and systems.
  5. Cost governance (FinOps). Whether spend is attributed to team, product or environment through tags, whether there are capacity commitments (reserved instances, savings plans) and when they were last reviewed, whether idle or oversized resources are found systematically.

The questions by role

Infrastructure architects. What is the adoption model and level of maturity? How is the infrastructure created? What is the network topology? How are privileged identities managed? What is the backup and recovery strategy? How are security baselines enforced (SCP, Azure Policy, Org Policy)?

DevOps and platform engineering. Which CI/CD tools? Where are the secrets? How are containers orchestrated and is the cluster hardened according to a document? How does observability work and do the alerts lead to an action? How often are images and systems updated?

Finance and FinOps. Is spend tracked by team, product, environment? Are there capacity commitments and when were they last reviewed? Who deals with unused resources?

The evidence

  • the current architecture diagram;
  • the structure of the IaC repository and the share of infrastructure it covers;
  • security scores (AWS Security Hub, Microsoft Secure Score, GCP Security Command Center);
  • the outcome of the review of IAM policies and privileged access;
  • the cost dashboard or FinOps report for the last quarter;
  • the report of the last disaster recovery test.

The warning signs

  • Resources created by hand, with no governance through IaC.
  • Long-lived credentials or access keys not rotated for a long time.
  • No MFA on console access.
  • Public buckets (S3, Azure blob, GCS) with no explicit justification.
  • No tagging strategy: cost attribution is impossible.
  • A disaster recovery test never run or more than a year old.

How to weigh it

Two criteria. The first: assess the findings against the pillars of the provider's Well-Architected Framework (security, reliability, cost, operational excellence, performance). The second: tell production environments from development ones, because the blast radius of a mistake is different. A misconfiguration in the cloud generally has a higher chance of being exploited than an equivalent one on-premises, because it can be reached from the internet: it must be weighed accordingly.

An example

A company discovers that its cloud bill has doubled in a year. The assessment does not start from prices but from tags: without tags nobody knows which product generates the spend. The first finding is therefore one of governance (no attribution), the second is a check of idle resources. Cost reduction comes after.

How to use the skill

The skill at the end brings these areas, questions and evidence in a form an agent can use, with or without interviews.

This article is an OIAR skill

What you have just read is the content of a skill from the OIAR catalogue, written for people. Knowledge like this, but richer and deeper, written for the agents, drives the interviews, the analysis and the report inside the platform. Skills are used in OIAR: to try this one on a project of yours you need an account.

cloud-assessmentKnowledge for assessing a cloud environment (AWS, Azure, GCP): adoption model, IaC, identity, network, backup and DR, CI/CD, observability, cost governance. Use it when the user asks for a cloud assessment, an architecture review or a pre-migration check, with or without interviews.